Preprint
English Accent Accuracy Analysis in a State-of-the-Art Automatic Speech Recognition System
Guillermo Cámbara et al., arXiv (2 pages, prepared for Phonetics and Phonology in Europe 2021), 2021
Read the originalRead on October 17, 2026
- What it is
- A short study that tested a large English speech model, trained on 44.5 thousand hours of English, against recordings of many English accents from the Common Voice corpus. It reports word error rate for each accent group.
- How much weight it carries
- A two-page conference paper on arXiv rather than a journal article. The per-accent results appear in a figure we could not read as text, so this library uses only the bias finding stated in the abstract, not the group-by-group numbers.
Facts taken from this source
- In a 2021 test of a large English speech model trained on 44.5K hours of English speech, the authors found an accuracy bias that favored the accents most prevalent in the training corpus.
“The model has been trained with 44.5K hours of English speech from an open access corpus called Multilingual LibriSpeech ... there is indeed an accuracy bias in terms of accentual variety, favoring the accents most prevalent in the training corpus” (Abstract)